Legionella, a genus of pathogenic Gram-negative bacteria, is widely present in natural water sources and artificial water systems. A total of 65 species and > 70 serogroups of Legionella have been characterized^[1]. Lipopolysaccharide (LPS), the main component of the outer membrane of Legionella, is not only associated with toxicity, but also provides the basis for classification in serotyping^[2]. LPS consists of lipid A, core polysaccharide, and O-antigen. O-antigens are glycopolymers expressed on the cell surface of Gram-negative bacteria. Variability in the O-antigen structure constitutes the basis for the establishment of serotyping. Based on variations in the O-antigen, Legionella can be divided into 15 serogroups^[3]. Approximately 84% of Legionella infections are caused by L. pneumophila serogroup 1 (LP1)^[4], and Legionnaires’ disease caused by non-L. pneumophila strains of Legionella accounts for approximately 5%–10% of all cases^[5].

LPS comprises 0–1.6% of the dry weight of L. pneumophila. LPS has been shown to stimulate morphological changes and alterations in gene expression in almost all host cells, leading to the uncontrolled expression of host cytokines, severe infection, and septic shock^[6]. Gene clusters encoding enzymes for the synthesis of Legionella LPS provide model candidates for studying molecular evolution at the DNA level. The Legionella O-antigen contains a number of isomers and derivatives of the monosaccharide pseudaminic acid, such as legionaminic acid and 4-epilegionaminic acid^[7]. This monosaccharide is rarely present in the O-antigen of bacteria, thus accounting for the unique O-antigen structure of Legionella.

In the present study, various Legionella serogroups were identified by PCR, and comparisons were made among the O-antigen-specific genes. The PCR method has been tested previously for its specificity and sensitivity^{[8, 9]}, and LPS composition may be a determinant of serogroup specificity, as defined by the immunofluorescence-based serotyping schema for L. pneumophila and other Legionella species^[2].

Whole genome sequences from 97 strains of Legionella were analyzed in the present study. Complete details of these strains, including the strain name, source, place of isolation, serotype, and time of isolation, are provided in Supplementary Table S1 available in www.besjournal.com. The data generated in this Whole Genome Shotgun project have been deposited in the National Center for Biotechnology Information (NCBI) under the BioProjectID PRJNA281151 with accession numbers LBAW00000000, LBHK00000000, LBAX00000000, LUCB00000000, LCUA00000000, LBAY00000000, LAVP00000000, and LBMS00000000.

The LPS synthesis gene sequences of all Legionella strains were downloaded in NCBI and multiple annotated databases (including NR, SwissProt, KEGG, COG, TCDB, go, PHI, VFDB, ARDB, Secretory-protein, T3SS, and CAZy). The LPS synthesis gene structure of each serogroup of Legionella was found to be similar. Among the LPS core genes of 15 serotypes (including serogroups LP1–14 and an unknown), the genomes of serotypes LP1, LP2, LP3, LP4, LP5, LP7, LP8, LP9, LP10, and LP13 contained 19 core genes. LP11 contained fewer LPS genes than LP1–LP5, LP7–LP10, and LP13, and lacked the genes lpg0837_YP_094872.1, lpg0838_YP_094873.1, lpg0920_YP_094954.1, lpg0749_YP_094785.1, lpg2630_YP_096635.1, lpxD_lpg2944_YP_096937.1, lpg2695_YP_096700.1, lpxB_lpg2945_YP_096938.1, lpg0748_YP_094784.1, and lpg2799_YP_096794.1 (Supplementary Figure S1, available in www.besjournal.com).

Figure S1.  Core gene LPS structure of Legionella pneumophila in each serogroup. The results obtained revealed that there are 15 LPS core genes in each serogroup. Each arrow represents one core gene. Identical genes are indicated using the same color.

Core-pan analysis of all serogroups (LP1–14 and the unknown) was then performed. Numbers of LPS core and pan genes, and the core-pan differences between each serogroup of Legionella, are shown in Supplementary Table S2 available in www.besjournal.com. For the pan genes of LPS, the following differences were observed: Among the 15 serotypes, LP1 and the unknown contained the lpg0777_YP_094813.1 gene (an O-acetyltransferase), and unlike the other Legionella types, LP4, LP5, LP8, LP9, LP11, and LP13 lacked the lpg0782_YP_094818.1 gene (encoding a putative O-acetyltransferase). Among the core genes for LPS synthesis, LP14 uniquely contained a gene, alpg2600_YP_ 096605.1, whereas LP11 contained fewer LPS pan genes compared with other serogroups of Legionella (Supplementary Figure S2, available in www.besjournal.com).

Figure S2.  Pan gene LPS structure of Legionella pneumophila each serogroup. The results obtained revealed that there are 15 LPS pan genes in each serogroup. Each arrow represents one pan gene. Identical genes are indicated using the same color.

The sequences of the core and pan genes of each serogroup were then compared with the previously downloaded reference sequences (in BLAST), and filtered based on the following criteria: identity > 80%, and coverage > 80%. Gene clusters were classified according to the LPS sequence length (small to large), and the identical LPS genes of different serogroups were connected by coarse lines. In addition, functional annotation of the LPS genes was performed for each serogroup. The LPS synthesis gene structure of non-L. pneumophila was found to be similar among non-L. pneumophila serogroups. However, among the LPS core genes, the following differences were observed: Legionella_dumoffii_NY_23, Legionella_cherrii_DSM_19213,Legionella_shakespearei_DSM_23087, Legionella_moravica_DSM19234, and Legionella_dumoffii_Tex_KL contained more LPS genes than other non-L. pneumophila serogroups, including lpxB_lpg2945_YP_096938.1 and lpxD_lpg2944_YP_096937.1. Furthermore, Legionella_dumoffii_NY_23, Legionella_dumoffii_Tex_KL, and Legionella_anisa_linanisettee contained the gene lpg0749_YP_094785.1, whereas other non-L. pneumophila serogroups did not (Supplementary Figure S3, available in www.besjournal.com).

Figure S3.  Core gene LPS structure of Legionella non-pneumophila. The results obtained revealed that there are 22 LPS core genes in each serogroup. Each arrow represents one core gene. Identical genes are indicated using the same color.

Among the 15 serogroups of L. pneumophila, LP 14 contained the largest number of LPS genes (21 core genes and 21 pan genes), and L. pneumophila serogroup 11 contained the fewest LPS genes (9 core genes and 11 pan genes). The genes held in common between L. pneumophila and non-L. pneumophila were found to be ostA_lpg0297_YP_094351.1, lpg0296_YP_094350.1, lpg0295_YP_094349.1, lpg2695_YP_096700.1, lpg2630_YP_096635.1, lpg2629_YP_096634.1, kdsA_lpg1182_YP_095215.1, pyrG_lpg1181_YP_095214.1, lpg0839_YP_094874.1, lpg0838_YP_094873.1, lpg0836_YP_094871.1, and CAB65217.1.

Among the non-L. pneumophila serogroups, 14–29 core LPS genes were identified, and these also had a higher distributional difference. L. pneumophila serogroups were found to contain a higher number of LPS synthesis genes compared with the non-L. pneumophila serogroups, and greater LPS gene differences were also detected among non-L. pneumophila serogroups, especially Legionella_dumoffii_NY_23,Legionella_dumoffii_Tex_KL, and Legionella_anisa_linanisettee (Supplementary Figures S2 and S3).

The LPS genes for all Legionella strains were downloaded from the NCBI database, and homologous alignment was subsequently performed for all samples (including 76 strains of L. pneumophila, 21 strains of non-L. pneumophila, and one strain of Coxiella burnetii). Screening samples with the LPS reference sequence annotation, the same information with the LPS reference sequence annotation was chosen. The above results were merged into redundancy, core-pan analysis was performed based on core gene similarity, and a phylogenetic tree was constructed using the neighbor-joining method. Based on the presence of pan genes, a phylogenetic tree was constructed using the unweighted pair-group method with arithmetic means. The phylogenetic tree was constructed according to the LP grouping system using the neighbor-joining method and unweighted pair-group method. After adding the external reference Rickettsia, the results obtained were similar to those without the external reference. The analyzed data revealed a state of gradual evolution of these strains from non-L. pneumophila to L. pneumophila. WZ1519012 (LP11), a strain of L. pneumophila, represented the closest association with non-L. pneumophila, although the distribution of different serogroups of Legionella species was dispersed. According to the phylogenetic tree constructed based on core gene similarity, the evolutionary edge strains included Legionella_shakespearei_DSM_23087_uid199258 and WD-4-1102. WD-4-1102 belongs to LP14 (Figure 1). In the phylogenetic tree based on pan gene similarity, the edge strains included SH003 and Legionella_moravica_DSM_19234_uid185635, although Legionella_shakespearei_DSM_23087_uid199258 and WD-4-1102 were not far apart (Figure 2).

Figure 1.  Phylogenetic tree based on core genes. Branch lengths are calculated from the means of the posterior probability density. Values below the nodes represent posterior probabilities. The scale bar represents substitutions per site.

Figure 2.  Phylogenetic tree based on pan genes. Branch lengths are calculated from the means of the posterior probability density. Values below the nodes represent posterior probabilities. The scale bar represents substitutions per site.

The difference identified between the two methods was that L. pneumophila_Paris_uid13127 (LP1) and L. pneumophila_Lens_uid13126 (LP1) were clustered together in the pan tree. Whole LPS genes of L. pneumophila_Paris_uid13127 (LP1) and L. pneumophila_Lens_uid13126 (LP1) were similar to those in other samples, although larger numbers of specific genes (264 and 223, respectively) were observed. Evolutionary trees with or without an external reference produced similar results. In the pan tree, SH003 represented the closest association with the evolutionary edge, whereas this was not the case in the core phylogenetic tree (Figure 1 and Figure 2).

The results of phylogenetic analysis of core genes showed that Legionella_dumoffii_NY_23,Legionella_dumoffii_Tex_KL, Legionella_cherrii_DSM_19213, and Legionella_anisa_linanisettee shared a close evolutionary relationship. Based on the phylogenetic analysis results, it was possible to speculate on how evolution may have progressed from non-L. pneumophila to L. pneumophila. WZ1519012 (LP11) represented the closest association with non-L. pneumophila; however, the distribution of serogroups of different Legionella species was more dispersed.

Primer sequences for LP11 and LP14 were tested and screened by PCR based on O-antigen-specific genes. PCR results for hypothetical genes were eliminated in initial screening, and Legionella serogroups 11 and 14 were chosen as the PCR templates. The O-antigen-specific genes for Legionella were repeatedly screened by PCR. Primers were designed according to the specific wzt gene in the gene cluster of Legionella serogroups 11 and 14. Using these primers, bands of the correct (expected) size were obtained after PCR. The PCR reaction mixture comprised, in a total volume of 30 μL the following: 15 μL of Premix Taq (TaKaRa Taq Version 2 plus dye; Takara Biotechnology Co., Ltd.), 1 μL of forward primer (10 μmol/L), 1 μL of reverse primer (10 μmol/L), 11 μL of distilled water (Invitrogen® UltraPure Distilled Water; Thermo Fisher Scientific, Inc.), and 2 μL of template (50 ng). The PCR thermocycling conditions were: 95 °C for 10 min, followed by 35 cycles of 95 °C for 30 s, annealing for 45 s (for the indicated temperatures, see Supplementary Table S3 available in www.besjournal.com), and 72 °C for 1 min, followed by 72 °C for 10 min. The primer sequences for LP11 and LP14, the size of the PCR products, and the annealing temperatures are shown in Supplementary Table S3. With the exception of the positive control group, none of the other groups presented with bands of the correct size. Hence, this demonstrated that the wzt gene was highly specific in Legionella serogroups 11 and 14.

The PCR method was tested for its specificity and sensitivity using agarose gel electrophoresis. The electrophoresis conditions were as follows: the gels were run for 40 min at 100 V in 0.5× TBE (Beijing Solarbio Science & Technology Co., Ltd). The results obtained revealed that LP11 (Figure 3A and 3C) and LP14 (Figure 3B and 3D) displayed bands for the target gene, whereas other serogroups did not. Therefore, the specificity and sensitivity of the primers of LP11 and LP14 were confirmed by agarose gel electrophoresis.

Figure 3.  Agarose gel electrophoresis of PCR products. (A) LP11 PCR. PCR using primers designed for LP11: lane M, 100-bp DNA marker; lanes 1–15, L. pneumophila LP1 to LP15; lane 16, negative control. (B) LP14 PCR. PCR using primers designed for LP14: lane M, 100-bp DNA marker; lanes 1–15, L. pneumophila LP1 to LP15; lane 16, negative control. (C) LP11 PCR. PCR using primers designed for LP11: lane M, 100-bp DNA marker; lane 1, Legionella gormanii_ATCC 33297; lane 2, Legionella dumoffii Tex-KL; lane 3, Legionella_cherrii_DSM_19213_uid222237; lane 4, Legionella longbeachae_ATCC33462; lane 5, positive control; lane 6, negative control. (D) LP14 PCR. PCR using primers designed for LP11: lane M, 100-bp DNA marker; lane 1, Legionella gormanii_ATCC 33297; lane 2, Legionella dumoffii Tex-KL; lane 3, Legionella_cherrii_DSM_19213_uid222237; lane 4, Legionella longbeachae_ATCC33462; lane 5, positive control; lane 6, negative control.

In conclusion, the present study has shown that, based on the wzt gene in the LPS cluster, it was possible to establish a rapid and specific method suitable for the identification of L. pneumophila serogroups. Application of this technique enabled identification of the genetic determinants of non-L. pneumophila virulence, and allowed for important comparative studies with other Legionella species to be made. Hence, gene chip- or PCR-based methods can be applied to detect different serogroups of Legionella easily and rapidly.